Four Japanese outlets report that NEC stopped building quantum machines in March. NEC has not confirmed the halt; a new Moonshot roster lists Fujitsu, not NEC.
Four Japanese outlets reported between 4 and 10 September that NEC stopped quantum machine development at the end of March 2026. NEC has not confirmed the reported halt. The published record consists of those reports, all relying on unnamed or unattributed sourcing, plus NEC statements that quantum research continues.
Diamond Online published first, at 20:00 JST on September 4, 2026, citing people familiar with the matter: NEC withdrew from machine development at the fiscal year-end, and many of the researchers involved moved to Fujitsu. Nikkei Asia followed on September 5, 2026, saying NEC "has halted development of a working quantum computer, it was learned Saturday, after apparently determining that development would likely take too long to reap a return on investment." Jiji Press reported the end-of-March halt on September 7, 2026, and described cost-effectiveness as the apparent judgment. Kyodo, on September 9, 2026, cited "sources familiar with the matter" as saying NEC determined the project "would require over a decade of further investment before it could be put to practical use and that it would be difficult to monetize."
Tom's Hardware reported that Diamond asked NEC whether it had exited quantum computer development; according to that English paraphrase of the paywalled Japanese article, NEC declined to comment on an exit and said it continued application, industrialization and customer proof-of-concept work. Kyodo carried a separate on-record line: "NEC would continue quantum research, a company official said." In checks conducted September 13, 2026, SCN found no halt statement in NEC's global or Japanese newsroom listings; the newest item visible on its Japanese quantum hub was dated April 11, 2025.
One primary document bears on the reports, and it predates them. JST's page for the superconducting project it selected for Phase 2 of Moonshot Goal 6, last updated August 31, 2026, includes a performer list. Fujitsu and RIKEN are on it. NEC is not.
The lineage the reports invoke starts with a letter in Nature on April 29, 1999. Yasunobu Nakamura and Jaw-Shen Tsai of NEC Fundamental Research Laboratories, with Yuri Pashkin, listed by Nature at CREST/JST, reported "the observation of quantum oscillations in a single-Cooper-pair box," a superconducting island whose two charge states they could place in superposition with a voltage pulse. It was the first coherent control of a superconducting solid-state qubit; Jiji and Kyodo both describe it as the source of the approach Google and IBM later built on. Nakamura is now at RIKEN and the University of Tokyo, Tsai at Tokyo University of Science. Neither is reported to be part of any 2026 move.
NEC's documented hardware work in the 2020s ran on two tracks. One was a superconducting-parametron annealer. The other, its Moonshot project, separately addressed gate-model superconducting integration and error correction. An annealer is a special-purpose device for combinatorial optimization; it settles coupled qubits into a low-energy state that encodes the answer and cannot run the general gate operations of a universal machine. NEC used superconducting parametrons, oscillator-based qubits that store information in the phase of a driven oscillation, rather than the transmons IBM and Google use. In February 2021, NEC licensed the ParityQC architecture and said it was "developing quantum annealers using superconducting parametron qubits ... with an aim to realize practical machines by 2023," with partial support from NEDO, Japan's New Energy and Industrial Technology Development Organization. In March 2022, it announced a four-qubit LHZ unit cell, a block meant to tile out into a fully connected annealer, and restated "the aim of realizing quantum annealing machines by 2023."
In June 2023, NEC announced an eight-qubit machine, developed with AIST and opened to Tohoku University for joint research. NEC's release called it "the first domestically manufactured quantum annealing machine in Japan that is accessible from the outside via the Internet." SCN found no later, larger annealer announcement in NEC's release archive through September 13, 2026, and could not verify from the reviewed NEC and Tohoku pages whether internet access remains available. Kyodo put the gap plainly: NEC "has lagged in recent years, having only made an eight-qubit machine, compared with machines with several hundred qubits of computing ability developed in other countries."
A separate NEC service change preceded the reported hardware halt. The company's service pages say its Vector Annealing cloud service, which ran on SX-Aurora TSUBASA vector supercomputers, ended in March 2025 ("2025年3月末をもって提供終了"). Vector Annealing 4.0 remains available as x86 software, alongside resale of D-Wave's Leap cloud service and optimization consulting.
Moonshot Goal 6, run by the Cabinet Office and JST, targets a fault-tolerant universal quantum computer by 2050, with a 2030 milestone to demonstrate error correction on "a certain scale of NISQ computer." NEC's Stage 1 project, "Development of Integration Technologies for Superconducting Quantum Circuits," ran from FY2020 to FY2025 under project manager Tsuyoshi Yamamoto, whom JST listed as a research fellow at NEC. Its performers included NEC, AIST, NTT, RIKEN and Tokyo University of Science.
The successor, "Development of Superconducting Fault-Tolerant Quantum Computer Systems," was selected for Phase 2 in FY2025 with the same manager, now listed as "Joint Appointed Fellow, National Institute of Advanced Industrial Science and Technology." The project page, which JST marks as last updated August 31, 2026, lists 25 named performers from 18 organizations: RIKEN, NTT, Tokyo University of Science, the University of Tokyo, AIST, Keio University, Fujitsu Limited, Kyocera Corporation, Tokai National Higher Education and Research System, Tohoku University, NanoBridge Semiconductor, the National Astronomical Observatory of Japan, NICT, the University of Osaka, Chuo University, Kyushu University, Ulvac Cryogenics and Ulvac. NEC does not appear. The 2028 milestone is a chip with more than 50 physical qubits and frequency tuning to suppress variation between them. By 2030, the target is "a fault-tolerant quantum computing system equipped with a quantum processor that achieves uniformity and high density for more than 100 physical qubits." The summary says "the number of physical qubits must be scaled up to roughly ten thousand times the current level" and proposes moving the wide-bandwidth control electronics into chiplets operating at cryogenic temperatures.
Two cautions apply. The manager's listing is a change in what JST records, not evidence of an employment transfer. And a page dated August 31, 2026 says nothing about when NEC left the project or why.
SCN has reported how Europe sorts quantum hardware by funding instrument as much as by qubit type; read that way, the Japanese record shows an instrument continuing with a changed roster, and nothing about what happened inside NEC.
Fujitsu and RIKEN unveiled a 256-qubit superconducting machine at the RIKEN RQC-FUJITSU Collaboration Center in Wako on April 22, 2025, four times the 64-qubit system of October 2023. Fujitsu said a 1,000-qubit computer was "scheduled to be installed in a new building at Fujitsu Technology Park in 2026" and extended the collaboration center's term to March 2029. On August 1, 2025, Fujitsu said it had started developing a superconducting computer "exceeding 10,000 qubits," operating 250 logical qubits, for completion in fiscal 2030, under a NEDO project with AIST and RIKEN that runs until fiscal 2027; it is targeting a 1,000-logical-qubit machine for fiscal 2035. That NEDO project is a separate instrument from the Moonshot consortium. As of September 13, 2026, SCN found no announcement that the 1,000-qubit system has been installed.
On September 8, 2026, the day before Kyodo's report, Fujitsu announced what it called "the world's first working prototype of a diamond-spin quantum computer" and said the approach "has the potential to be integrated with superconducting quantum computers." Kyodo quoted Shintaro Sato, head of quantum research at Fujitsu: "We require further governmental support."
On the researchers: Diamond reported that many moved to Fujitsu, including one it called an ace. Diamond's September 9, 2026 follow-up quoted Fujitsu's head of quantum development as saying it would continue strengthening its researcher ranks. Neither company has named an individual, and neither has confirmed a transfer.
The reasoning attributed to NEC, that practical use is more than a decade away and hard to monetize, is a corporate judgment about its own program. It differs from public roadmaps that target milestones between 2028 and FY2030. IBM's May 2026 SEC filing disclosed a $10 billion five-year quantum plan, which SCN reported alongside IBM's 2029 target for Starling, a roughly 200-logical-qubit system. The US Department of Energy's request for information specifies logical qubits in the low hundreds, and SCN has reported that its 2028 aim is hard to define. Fujitsu's own figure is 250 logical qubits at fiscal 2030, and the Moonshot Phase 2 milestone is more than 100 physical qubits by the same year. Those roadmaps describe different systems and milestones, and none contradicts NEC's assessment of its own program.
The sovereignty reading is narrower. A June 24, 2026 Cabinet Office table projects ¥10.3 trillion in public and private investment in quantum computing through FY2040, one line among 62 in the roadmap for the government's 17 strategic fields. The Institute for Molecular Science's Shunkai neutral-atom system targets 10,000 physical qubits by 2031, Hitachi is in Phase 2 on silicon qubits, and Fujitsu now has a diamond-spin prototype beside its superconducting line. Japan's largest working machines are superconducting. The new roster puts Fujitsu inside Japan's Phase 2 superconducting effort, while NEC is absent from the published performer list; it does not reduce that effort to Fujitsu and RIKEN alone.
The SCN Staff is a small AI editorial squad working under human direction. Each agent owns one job.
Scout does the research. It runs down primary sources and checks what's already been published, on SCN and everywhere else, before a story gets written. If a claim can't be traced back to a real document, Scout flags it.
Forge writes. It takes what Scout found and turns it into a draft, argument and sentences and all. Every SCN piece starts here, then gets sharpened.
Cipher handles search: the titles, descriptions, and keyphrase work that decides whether a good article ever gets found. Least glamorous job on the squad. Also one that matters more than it looks.
Pixel makes the visuals. Images, charts, the occasional diagram, all built to SCN's brand instead of pulled from a stock library. When something's easier to see than to read, it goes to Pixel.
Editorial judgment and the final call stay with the humans. So does the fact-checking.
